PAGE 1    PAGE 2

DIY CONTROL LOADER YOKE PROJECT

I've been working on the development of a DIY control loading flight yoke, this is a continuation of the force feedback joystick project I experimented with before. The project has been through a number of development iterations and although I'm still not 100% happy with the end result I thought it would be worth a write-up at the moment while I figure out further improvements. It is a fascinating project and has good potential.

My intention was to make a working force feedback flight yoke that had some of the features of commercial control loader yokes but without the huge price tag usually attached to them.

I think I've managed to do this - but not well enough yet to my satisfaction. Although I think the basic yoke works well the main problem that remains is the smoothness of the torque output from the drive motors and the effect on the smoothness of the yoke loading. Whereas I've managed to make significant improvements in force smoothness through motor choice and control hardware and software design improvements, light but significant force fluctuations still remain in the load response of the yoke and these can be detected by touch. In my view they are still noticeable enough to detract from the force feedback effect.

There are solutions to this, but keeping the cost of those solutions down is tricky - more on this later.

QUICK REVIEW OF DEVELOPMENTS

All the yoke designs I've worked on use electric motors to generate the loading on the elevator and aileron axes. I've experimented with stepper motors, conventional low cost DC motors, higher spec "low-cogging" DC motors, brushless motors and finally skewed rotor DC servo motors.  The last two - brushless and skewed rotor provide the best results so far with the skewed rotor edging it on the basis of robustness and system cost. However I'm not entirely happy with the results yet and it is likely that further trials with ironless disc-armature "pancake" motors will be needed to get the final improvements I'm looking for.

In addition to the drive motor developments I've also experimented with various software and hardware improvements. Control loading software I've written for the system drives the force demand to the aileron and elevator axes of the yoke. Control surface position, speed and acceleration are determined from the yoke movements, flight data such as airspeed, aircraft accelerations, stall conditions, gear position, engine speed etc is extracted from MSFS and instantaneous yoke forces are calculated and sent to the control hardware. Improvements to the software structure and position reporting hardware allow this to be done every 12 ms or so (80Hz) which significantly improves the fidelity and stability of the control system over my previous attempts.

Trials on the motor control hardware have included Dimension Engineering Sabertooth 2x25 and Devantech MD03 DC motor controllers and a sinusoidal commutation / hall effect feedback driver for  the brushless motor. The behaviour of different DC controllers can vary significantly in terms of output resolution and zero-position null bands and transition behaviours. With brushless controllers it seems it is the commutation method that is significant in terms of its effect on torque smoothness.

YOKE MECHANISM & DRIVE MOTORS

The latest prototype force feedback yoke is shown in the images. The aileron axis rotary mechanism is mounted on a linear movement carriage which provides the elevator movement. Both axes are connected to an electric motor by toothed belt drives. The toothed belts are a convenient means of force transmission for this design as they allow some flexibility in component positioning, zero backlash, no cogging and a choice of reduction ratio.

The main yoke column is roller bearing mounted to provide low friction rotation. The linear carriage is carried via ball bushings on two steel guide rods. Overall there is very little friction in the mechanism and this contributes to the smoothness essential for the application. The linear guide system is also fairly stiff and would allow for the yoke column to be extended to suit customised installation in a cockpit. At present the yoke itself is a simple plywood profile fixed to the hollow tube end - a more sophisticated yoke could be added with its button cabling running through the hollow support column.

The aileron axis movement is not currently mechanically limited in the design - I have it electrically calibrated to +/- 90 Deg yoke rotation. The elevator axis travel is about 130mm but could be extended without difficulty.

The choice of suitable electric motors has been a major challenge in the project especially considering the desire for a low cost design; the present motors as shown in the above photos are skewed rotor DC PM servo motors which I run on a 12V system. Their rated torque output is 1.2 Nm with peaks up to 6 Nm and these outputs are sufficient to generate strong yoke forces. The whole yoke assembly has to be clamped to a work bench to ensure it doesn't move when the control forces come on.

A major issue driving motor choice is torque smoothness. All the low cost options I have tried - including stepper, low cost PMDC, and "low cogging" PMDC servo motors (see left) - have resulted in transmission through to the yoke of cogging torque detents which ruin the force feedback feel. A brushless motor drive proved more successful, however even here a position dependent torque "detent" arising from the brushless driver's commutation method was transmitted to the yoke. Unlike the previous motors' cogging torque which was broadly fixed in magnitude this commutation related detent torque was current related - the higher the load the more noticeable the detents were, although it was fine at low loads. Additional "gearing" in these systems does not help this situation - just as gearing magnifies the base torque output it also magnifies the torque variations. What is more important is the ratio of any torque variation to the base torque output of the motor - this must be very low.

The brushless drive I tried was one of the least expensive I could find and it is therefore not the smoothest brushless drive type available. However the cost of higher performance AC brushless drives which use high resolution external position encoders in addition to the motor's internal hall-effect position feedback is in my view simply too high for a DIY flight yoke.

The most recent motor type I have tested is a skewed rotor PMDC servo motor. The skewed rotor construction effectively eliminates the cogging torque variations and the motors turn smoothly without significant torque variation at low loads. However they are still brushed DC motors and use mechanical commutation with brushes. The effect of this is that the low speed torque output smoothness is subject to how the brushes energise adjacent rotor windings and this is dependent on the quality of the brush/commutator contact. Smooth overlap of the brushes over adjacent commutator segments is needed to reduce the dip in torque between winding energising and poorly bedded-in brushes proved to be a problem here.

Dressing the brushes to increase the effective contact area helped, however even then there is still a torque ripple which comes through due to the commutation mechanism and the ripple is proportional to motor current - as with the brushless drive.

I've gone with the skewed rotor PMDC motors for now as the final result is much much better than previous types but I still don't think it's quite good enough given the cost of the DIY yoke - there's more testing to come with other motor types! Ironless disc-armature pancake style motors have zero cogging and many more commutator segments than the more conventional DC motor types I've used. They offer perhaps the best option for a smooth torque output whilst retaining low-cost DC drive hardware.